Thirty probes.
One consensus.
Nine check types, multi-region quorum, eight alert integrations, maintenance windows, and Playwright synthetic checks -- from one pane.
Nine ways to
prove it's up.
From simple HTTP status codes to full Playwright browser scripts. Each check type has its own assertion language and failure semantics.
| Type | What it checks | Port |
|---|---|---|
| HTTP | Status code, body match, response time | 80/443 |
| TCP | Connection established within timeout | any |
| DNS | Query response matches expected answer | 53 |
| ICMP | Ping round-trip time and packet loss | - |
| SSL | Certificate validity, expiry, chain trust | 443 |
| Keyword | HTTP body contains or excludes string | 80/443 |
| Heartbeat | Expect periodic POST from your cron | inbound |
| Playwright | Full browser script, screenshots on fail | headless |
| UDP | Send payload, validate response bytes | any |
One probe fails.
No false alarm.
Relays runs every check from 30 probes in 8+ regions. An incident triggers only when a configurable quorum (default 60%) confirms failure. One flaky probe never pages your team at 3am.
| Region | Location | Result | Latency |
|---|---|---|---|
| IAD | Virginia | … | … |
| AMS | Amsterdam | … | … |
| NRT | Tokyo | … | … |
| SYD | Sydney | … | … |
| GRU | Sao Paulo | … | … |
| FRA | Frankfurt | … | … |
| SIN | Singapore | … | … |
| ORD | Chicago | … | … |
Eight ways to
get paged.
Route alerts by severity, component, and time of day. Each channel supports auto-resolve on recovery so you never chase stale alerts.
Per-incident digest with timeline
Rich blocks with ack/resolve buttons
Events API v2, auto-resolve on recovery
Priority mapping, team routing
Signed JSON POST with retry + DLQ
Twilio-backed, 195 countries
Embed with status color + sparkline
Adaptive card with action buttons
Deploy without
false alerts.
Create time-boxed maintenance windows that suppress alerts for specific monitors. Subscribers are notified before, during, and after.
Real browser.
Real checkout.
Write Playwright scripts that run on a schedule from multiple regions. If the script fails, Relays captures a screenshot and creates an incident automatically.